Predictors of Treatment Response in Chronic Hepatitis B
- First Online:
- Cite this article as:
- Wong, G.L. & Chan, H.L. Drugs (2009) 69: 2167. doi:10.2165/11319850-000000000-00000
- 122 Views
The ultimate goal of treatment for chronic hepatitis B is to reduce liver-related complications and mortality. Sustained hepatitis B e antigen (HBeAg) seroconversion and hepatitis B surface antigen (HBsAg) clearance 6–12 months after stopping treatment are the short-term surrogate outcomes for interferon or peginterferon therapy. As most patients require long-term nucleos(t)ide analogue treatment, which also has the risk of drug resistance in the case of incomplete viral suppression, maintained hepatitis B virus (HBV) DNA suppression to an undetectable level is the appropriate surrogate outcome. Because no antiviral treatment is perfect, it is desirable for treatment response to be predicted and the treatment regimen modified accordingly. At baseline, high ALT and low HBV DNA levels can predict response to both (peg)interferon and nucleos(t)ide analogues. Genotype A HBV responds best to peginterferon but HBV genotype has no predictive value for nucleos(t)ide analogue treatment. HBV DNA is a good on-treatment predictor of response for nucleos(t)ide analogues but not for (peg)interferon. The data supporting the use of quantitative HBsAg and HBeAg to predict response to peginterferon is stronger than that for nucleos(t)ide analogues. In conclusion, predictors of response are useful to provide the most appropriate antiviral therapy to the most suitable patients, in order to achieve the best response and improve the clinical outcome of chronic hepatitis B patients.
Hepatitis B virus (HBV) infection is a global health problem, currently affecting about 400 million people worldwide. It is estimated that over 200 000 and 300 000 chronic hepatitis B patients die annually worldwide from liver cirrhosis and hepatocellular carcinoma (HCC), respectively. Persistent viral replication is associated with progression of liver disease. Hence, antiviral therapy is aimed at maximal viral suppression to achieve improved clinical outcomes. As no antiviral treatment is perfect, it is desirable to be able to predict treatment response and to modify the treatment regimen accordingly. As liver-related complications take decades to develop, most treatment trials use short-term surrogate markers of response as outcome measures. To evaluate predictors of treatment response, one must understand the meaning and importance of each of these surrogate markers. Furthermore, the on-treatment response of nucleos(t)ide analogues must be clearly differentiated from the off-treatment response of interferon (or peginterferon) therapy.
1. Surrogate Markers for Long-Term Outcome
1.1 Hepatitis B Surface Antigen (HBsAg) Clearance
Hepatitis B surface antigen (HBsAg) clearance, with or without the appearance of antibody to HBsAg (anti-HBs), is considered the criterion for resolution of HBV infection and probably the best definition of virological response.[3–5] Longitudinal studies have demonstrated that HBsAg seroclearance usually confers excellent long-term prognosis, provided that HBsAg loss occurs at a young age (<50 years), and precedes the development of advanced liver fibrosis or cirrhosis. Unfortunately, this is rarely achieved either spontaneously or by antiviral treatment. HBsAg clearance is usually a sustained event unless the patient becomes immunocompromised. However, HBsAg has been reported to reappear after the cessation of peginterferon treatment despite continued adefovir therapy in a small German series.
1.2 Hepatitis B e Antigen (HBeAg) Seroconversion
In patients with hepatitis B e antigen (HBeAg)-positive chronic hepatitis B, clearance of HBeAg and seroconversion to positive antibody to HBeAg (anti-HBe) is usually regarded as a desirable outcome of antiviral therapy.[3–5] Clearance of HBeAg, whether spontaneous or after antiviral therapy, reduces the risk of hepatic decompensation and improves survival.[8–11] Nonetheless, 10–30% of patients may still have elevated ALT and high HBV DNA levels after HBeAg seroconversion, and 10–20% of inactive carriers may have reactivation of HBV replication and exacerbations of hepatitis after years of quiescence.[12–14] Most hepatitis reactivation post-HBeAg seroconversion occurs within 6 months after stopping interferon treatment, but it may occur much later, up to a few years after the cessation of lamivudine.[15,16] Therefore, HBeAg seroconversion can serve as a surrogate marker of response only if it is sustained for at least 6 months after stopping treatment, and it may be a better surrogate marker for (peg)interferon than nucleos(t)ide analogues.
1.3 Hepatitis B Virus (HBV) DNA Suppression
Several large prospective Asian studies found that a high level of HBV DNA was an independent risk factor for the development of liver cirrhosis and HCC.[17,18] Hence, persistent suppression of HBV DNA, ideally to an undetectable level but at least to below 10 000 copies/mL, is another acceptable surrogate marker of treatment response. In studies of lamivudine treatment for patients with HBeAg-negative disease or advanced liver fibrosis, maintained viral suppression was associated with reduced cirrhotic complications and HCC.[11,19,20] A recent meta-analysis found the risk of HCC can be greatly reduced with viral suppression by antiviral therapy. However, viral relapse after cessation of antiviral treatment may negate the previous benefit. After stopping nucleos(t)ide analogues, viral relapse will occur in most HBeAg-positive patients who do not lose HBeAg as well as most HBeAg-negative patients. The durability of drug-induced viral suppression after 1 year of lamivudine or adefovir dipivoxil treatment in HBeAg-negative patients has been less than 10%.[3,23] For HBeAg-negative patients receiving peginterferon, most who sustain HBV DNA suppression to below 10 000 copies/mL for 1 year post-treatment will have sustained viral suppression until the third year. Therefore, the use of HBV DNA levels as a surrogate marker for response is valid only if viral suppression is maintained during nucleos(t)ide analogue treatment. For peginterferon treatment of HBeAg-negative patients, sustained HBV DNA suppression till 1 year post-treatment is also a reasonable surrogate marker of response.
1.4 Drug Resistance
One peculiar feature of the use of nucleos(t)ide analogues is the possibility of drug resistance. The development of drug resistance may lead to virological breakthrough, biochemical flare-up, histological deterioration, hepatic decompensation and increased risk of hepatic complications.[11,25,26] Although salvage therapies are available for drug resistance, it is not always possible to achieve good viral suppression.[27,28] Therefore, instead of monitoring for emergence of drug resistance, it is a good idea to identify patients who have a higher risk of developing resistance during treatment and modify the treatment regimen as early as possible.
2. Different Outcome Measures for (Peg)Interferon and Nucleos(t)ide Analogues
2.1 Sustained versus Maintained Response
Interferon and nucleos(t)ide analogues have different mechanisms of action. Interferon, either conventional or pegylated, has a predominantly immunomodulatory effect and weak antiviral activity. These dual actions result in suppression of viral replication and clearance of infected hepatocytes. Nucleos(t)ide analogues have a potent antiviral effect by inhibiting the reverse transcriptase and DNA polymerase activity of HBV. Because of these intrinsic differences between the two drug classes, interferon can be given for a finite duration, yet has a more durable off-treatment response. In contrast, nucleos(t)ide analogues are usually given for a prolonged duration and are associated with a less durable off-treatment response even after HBeAg seroconversion or adequate HBV DNA suppression. Therefore, different timepoints are usually used to assess the treatment outcome for these two different therapies.
Sustained response, which is defined as treatment response persisting for a certain duration after discontinuation of therapy, is more appropriate for interferon or peginterferon treatment. The off-treatment duration is usually at least 6–12 months and sometimes longer. Maintained response, which is defined as treatment response persisting throughout the course of treatment until the time of assessment, is more appropriate for nucleos(t)ide analogues. For maintained response to be meaningful, it is preferred that the duration of treatment is longer, probably up to a few years, particularly for drugs with a low genetic barrier of resistance. If sustained response is to be assessed for nucleos(t)ide analogues, a longer duration of off-treatment observation is needed, as late relapses after the first year post-treatment may occur.
2.2 Surrogate Outcome for (Peg)Interferon
The most commonly used surrogate outcome measure for interferon or peginterferon treatment is sustained HBeAg seroconversion, which is usually assessed 6–12 months after stopping treatment. In a long-term follow-up of a European multicentre study, 81% of the initial responders (HBeAg loss 6 months post-treatment) remained HBeAg-negative and 58% of the initial responders had HBV DNA levels lower than 10 000 copies/mL at 3 years post-treatment. Similarly, in a study in Hong Kong, 46% of the initial responders remained HBeAg-negative and 29% had HBV DNA levels lower than 100 000 copies/mL at 3 years post-treatment.
Sustained HBsAg clearance is a marker for more complete viral clearance than HBeAg seroconversion. However, HBsAg clearance is usually delayed. Studies in Europe and the US reported that delayed clearance of HBsAg occurred in 12–65% of patients within 5 years of HBeAg loss, but delayed HBsAg clearance was rarely observed in Asian patients.[8,31–35] Among patients who received 1 year of peginterferon treatment, 9–11% developed HBsAg clearance 3 years after treatment stopped.[24,30] Therefore, on-treatment predictors for HBsAg clearance are more valuable than those for HBeAg seroconversion, but much longer post-treatment follow-up is necessary for the evaluation of these predictors.
2.3 Surrogate Outcome for Nucleos(t)ide Analogues
Maintained viral suppression is the most commonly used surrogate outcome measure of efficacy for nucleos(t)ide analogues. As viral resistance may develop with ongoing viral replication activity, undetectable HBV DNA by polymerase chain reaction (PCR) is needed to prevent development of drug resistance. In other words, although there is no evidence to suggest that undetectable HBV DNA levels are needed for a better prognosis, HBV DNA levels at the cut-off of 10 000 copies/mL cannot be used as the surrogate marker in nucleos(t)ide analogue treatment. As HBsAg clearance is rarely observed with nucleos(t)ide analogues, evaluation of predictors for HBsAg clearance may be difficult and less clinically useful.[36–41]
At 1 year of nucleos(t)ide analogue therapy, the rates of achieving undetectable HBV DNA by PCR were reported in the range of 21–67% in HBeAg-positive patients and 51–90% in HBeAg-negative patients. For drugs with a low genetic barrier of resistance, new cases of drug resistance may emerge probably up to the fifth year of treatment.[37,42] Therefore, predictors of maintained HBV DNA response for these agents are more meaningful if they can predict the response at the end of 5 years. On the other hand, the rate of undetectable HBV DNA levels increased from 81% to 87% when the duration of entecavir therapy was increased from 48 to 144 weeks. Therefore, predictors of maintained HBV DNA response at a much shorter duration of treatment may be meaningful for drugs with a higher genetic barrier of resistance.
3. Predictors of Response to (Peg)Interferon
One-year of peginterferon treatment has become one of the first-line treatments of chronic hepatitis B.[3–5] Despite its advantage of a finite treatment duration and absence of drug resistance, peginterferon treatment is limited by its adverse effects, the need for subcutaneous administration and its relatively high cost. It would hence be desirable for peginterferon to be prescribed only to patients who have a high chance of response, whereas nonresponders could be identified so that peginterferon could be avoided or stopped early during the course of therapy. From the experience of chronic hepatitis C, a combination of baseline predictors and on-treatment virological response can formulate an algorithm for individualized treatment.
3.1 Baseline Predictors of Response to (Peg)Interferon
Based on the results of various well designed, randomized studies, patients who have good immune clearance tend to respond better to interferon. Immune clearance is signified by elevated ALT levels, lower HBV DNA levels and higher necroinflammatory activity on histology. High pretreatment serum ALT levels and low HBV DNA levels are associated with increased rates of HBeAg seroconversion in both conventional interferon- and peginterferon-treated patients.[45,46,48,49] High baseline ALT levels and low HBV DNA levels were also associated with response to peginterferon-α-2a in HBeAg-negative patients; ALT above 5 times the upper limit of normal (ULN) was the strongest predictor of combined response (ALT normalization, HBV DNA suppression) at 24 weeks post-treatment.
Quantitative HBsAg has recently become a hot area of research. Several studies have suggested a correlation of HBsAg titre with the level of covalently closed circular (ccc)DNA and total intrahepatic HBV DNA.[57,62] In a three-arm pilot study using different regimens of peginterferon and lamivudine in combination, patients with a lower cccDNA at baseline tended to have a higher sustained virological response (SVR).[57,58] This observation seems logical, as the lower cccDNA should be the result of vigorous host immune clearance. Furthermore, quantitative HBsAg, which has a moderate correlation with cccDNA, could provide a reasonable prediction of SVR at a level of <10 000 IU/mL. In another small study from the Netherlands, baseline HBsAg ≤2.25 log IU/L was found to predict HBsAg clearance in patients who received 48 weeks of peginterferon-α-2a in combination with adefovir dipivoxil. However, in the larger global peginterferon-α-2a study among HBeAg-positive patients, baseline HBsAg titre did not provide a good prediction for HBeAg seroconversion at 6 and 12 months after treatment cessation.
3.2 On-Treatment Predictors of Response to (Peg)Interferon
3.2.1 Serum HBV DNA
One way of using serum HBV DNA as an on-treatment predictor of response is to set an early stopping rule. In a post hoc combined analysis of two clinical trials using peginterferon-α-2b and lamivudine in combination in HBeAg-positive chronic hepatitis B,[58,66] failure of HBV DNA reduction to 10 000 copies/mL as early as week 8 has a negative predictive value of 92% for SVR. In the database of the global randomized study using peginterferon-α-2a monotherapy, a much higher HBV DNA cut-off of 9 log copies/mL at week 24 was found to have a negative predictive value of 86% for sustained HBeAg seroconversion. Whether the difference of these two studies is related to the nature of the peginterferon requires further study. Nonetheless, early HBV DNA kinetics may assist the clinical decision of stopping treatment particularly among patients who cannot tolerate the treatment well.
3.2.2 Quantitative HBsAg
Several studies have shown that end of treatment cccDNA and HBsAg levels are associated with SVR to peginterferon-based therapy. In a post hoc analysis of 47 patients after antiviral therapy, a lower cccDNA at the end of treatment was found to associate with a higher rate of SVR.[66,78] In 26 HBeAg-positive patients who completed a combination of 32 weeks of peginterferon-α-2b and 2 years of lamivudine treatment, an HBsAg level <1500 IU/mL has a sensitivity of 71% and specificity of 84% to predict SVR. In the global peginterferon-α-2a trial among HBeAg-negative patients, HBsAg level <10 IU/mL at week 48 and on-treatment decline >1 log10 IU/mL could predict sustained HBsAg clearance 3 years after peginterferon (with or without lamivudine) treatment.
As the serum HBsAg titre can correlate with cccDNA level, monitoring of serum HBsAg titre becomes a reasonable tool for on-treatment prediction for peginterferon therapy. In the global peginterferon-α-2a trial among HBeAg-positive patients, an HBsAg level decrease below 1500 IU/mL at week 12 or 24 could predict an increased rate (about 50%) of HBeAg seroconversion. Similarly, 35% of HBeAg-negative patients who could achieve HBsAg levels of 1500 IU/mL or less at week 12 of peginterferon treatment had HBsAg seroclearance by 4 years after peginterferon-α-2a with or without lamivudine. Recently, there has been some controversy about whether the absolute HBsAg level or reduction in HBsAg titre during treatment serves as a better predictor of response in HBeAg-negative patients. In the French study by Moucari and colleagues, a reduction in serum HBsAg by ≥1 log10 IU/mL at week 24 was found to have a negative predictive value of 97% for SVR to 48-week treatment with peginterferon-α-2a (undetectable serum HBV DNA levels 24 weeks post-treatment).
3.2.3 Quantitative HBeAg
Measurement of on-treatment HBeAg levels may provide additional information to predict the chance of HBeAg seroconversion to peginterferon-α-2a therapy. On analysing the 271 HBeAg-positive patients who received peginterferon-α-2a monotherapy in the global study, high levels of HBeAg (>100 Paul Ehrlich Institute Unit/mL) had a greater negative predictive value (96%) of SVR than that obtained for serum HBV DNA levels (86%). Hence, a high HBeAg level can be another criterion to predict nonresponders for early termination of treatment. Unfortunately, HBeAg titre quantification is not yet standardized, which hinders its application in clinical practice. In addition, HBeAg quantification is not useful in HBeAg-negative patients.
4. Predictor of Response to Nucleos(t)ide Analogues
Because most patients receiving nucleos(t)ide analogues require long-term treatment, careful assessment is needed before commencement of therapy. Predictors of response will be useful to assist the selection of patients who are more likely to respond. Furthermore, predictors for good responders, among whom nucleos(t)ide analogue therapy can be stopped after a finite duration without relapse, will also be useful.
A major concern with long-term nucleos(t)ide analogue treatment is the selection of drug-resistant mutations. The rate at which resistant mutants are selected is related to pretreatment serum HBV DNA level, rapidity of viral suppression, duration of treatment and prior exposure to nucleos(t)ide analogue therapies. According to American and European recommendations, nucleos(t)ide analogues with a high genetic barrier of resistance (entecavir and tenofovir disoproxil fumarate) should be used as first-line agents.[3,4] However, the high cost of these drugs is a major concern in Asian countries, where the relatively inexpensive antiviral agents with a lower genetic barrier of resistance (lamivudine, adefovir dipivoxil and telbivudine) are also accepted as first-line agents. In this scenario, prediction of response becomes important, as one may need to modify the treatment regimen among patients who have a high risk of developing drug resistance. Even when using drugs with a high genetic barrier of resistance, it may be useful to predict the potential nonresponders for closer surveillance of resistance and possible intensification of treatment.
4.1 Baseline Predictors of Response to Nucleos(t)ide Analogues
Table I presents baseline predictors of response to nucleos(t)ide analogues. Stronger host immune clearance, as reflected by a lower serum HBV DNA and higher ALT levels, can generally facilitate viral suppression by nucleos(t)ide analogues. Based on the analysis of databases in the pivotal trials, a lower baseline HBV DNA level is associated with a higher chance of HBeAg loss or seroconversion in HBeAg-positive patients receiving lamivudine or telbivudine.[51–53,56] A low baseline HBV DNA level was associated with maintained viral suppression at week 48 in adefovir-treated patients. On the other hand, most studies suggested that virological response (defined as HBeAg loss or seroconversion and/or undetectable HBV DNA) was not related to the HBV genotype.[41,56,81,82]
The rate of HBeAg loss was particularly high in lamivudine-treated patients when ALT level was higher than 5 × the ULN.[51,52] In a 4-year case-control study of lamivudine therapy in HBeAg-positive patients, HBeAg seroconversion was observed in 78% of patients with severe acute exacerbation of chronic hepatitis B (ALT >10 × the ULN with jaundice) versus 52% in the control group with less active disease. Nonetheless, the higher rate of HBeAg seroconversion could not be extrapolated to a higher rate of successful treatment cessation, as virological breakthrough (69%) and hepatitis reactivation (63%) were common even after HBeAg seroconversion in patients with severe acute exacerbation. In the telbivudine registration trial (GLOBE study), ALT level >2 × the ULN was predictive of HBeAg seroconversion. However, histological improvement or virological response was not associated with baseline ALT levels in adefovir dipivoxil- and entecavir-treated patients.[39,80] In HBeAg-negative patients with severe acute exacerbation receiving lamivudine, the cumulative probability of drug resistance seemed to be lower than that reported among patients with milder disease. However, in most major clinical trials, baseline ALT levels had a less pronounced or no effect on virological responses among HBeAg-negative chronic hepatitis B irrespective of nucleos(t)ide analogue studied.[56,80,83] In the phase III trial of tenofovir among HBeAg-positive patients for 48 weeks, patients with high baseline HBV DNA and HBsAg levels tended to have HBsAg loss. As the number of patients with HBsAg loss was very small, further studies are needed to confirm these findings.
4.2 On-Treatment Predictors of Response to Nucleos(t)ide Analogues
4.2.1 HBV DNA
On-treatment predictors of response to nucleos(t)ide analogues are presented in table II. The association of early viral suppression and the risk of drug resistance to different nucleos(t)ide analogues has been addressed in several studies. Greater HBV DNA reduction at 12 or 24 weeks was associated with reduced risk for subsequent lamivudine resistance.[67,68,84] In the GLOBE study, which compared the efficacy of telbivudine with lamivudine in both HBeAg-positive and HBeAg-negative chronic hepatitis B, complete viral suppression at 24 weeks of therapy could predict reduced risk of subsequent resistance in patients in both treatment arms.[56,69] Patients receiving lamivudine who could achieve undetectable HBV DNA levels at week 24 had only a 2–3% risk of resistance at year 1 and a 5–9% risk of resistance at year 2. Patients receiving telbivudine who could achieve undetectable HBV DNA levels at week 24 had 0–1% risk of resistance at year 1 and 2–4% risk of resistance at year 2. Adefovir dipivoxil has slower HBV DNA suppression than lamivudine and telbivudine, and week 48 is a more appropriate timepoint to predict its effectiveness. In a long-term follow-up study of HBeAg-negative patients on continuous adefovir dipivoxil therapy, patients who could achieve HBV DNA suppression to below 1000 copies/mL at week 48 had only a 6% risk of resistance over 192 weeks. Data on the use of early HBV DNA suppression to predict drug resistance or long-term response to entecavir and tenofovir disoproxil fumarate are lacking.
As HBV DNA suppression at weeks 24 to 48 can predict the subsequent risk of drug resistance, a roadmap model with on-treatment HBV DNA monitoring has been proposed. In this model, measurement of HBV DNA at week 24 was considered essential to characterize the early virological responses as complete (undetectable HBV DNA), partial (HBV DNA <2000 IU/mL or 4 log10 copies/mL) or inadequate (HBV DNA levels ≥2000 IU/mL or 4 log10 copies/mL). Patients with a complete early virological response are recommended to continue therapy with the same drug with regular monitoring of HBV DNA levels. Patients with partial or inadequate early virological responses may need to change to or add another antiviral agent(s), and the strategies would depend on the genetic barrier of resistance of the current nucleos(t)ide analogue. The aim of the roadmap model is to identify patients who have higher risk of drug resistance for pre-emptive intensification of treatment, while patients who have a lower risk of drug resistance can remain on the original treatment. It is particularly relevant when drugs with a lower genetic barrier of resistance are used.
4.2.2 Quantitative HBsAg
The evidence concerning the predictive value of quantitative HBsAg in patients receiving nucleos(t)ide analogues is evolving. One small study involving 21 HBeAg-negative patients receiving lamivudine found that the HBsAg level was reduced but at a significantly slower rate than with interferon (n = 42). The median estimated time to HBsAg undetectability derived from best curve fitting was 127 months for lamivudine virological responders and 65 months for interferon sustained responders. Another study involving 31 HBeAg-positive patients showed that HBsAg levels were significantly decreased during the first 12 weeks of adefovir dipivoxil with median change of −397.0 IU/mL, but no further HBsAg reduction was noted beyond 12 weeks. A recent analysis of the GLOBE study database showed that rapid on-treatment HBsAg decline of >1 log10 IU/mL during the first year of telbivudine is highly predictive of future HBsAg clearance. The clinical significance of a reduction in HBsAg level to predict virological response is yet to be determined. As the data on the use of quantitative HBsAg to predict response for nucleos(t)ide analogues are very preliminary, it cannot be recommended for clinical use at this stage.
Prediction of response to antiviral treatment has become increasingly important in the individualization of antiviral therapy for chronic hepatitis B patients. Baseline predictors can assist physicians to select the best timing for treatment initiation as well as the choice of antiviral agents ([peg]interferon vs nucleos(t)ide analogues). On-treatment prediction with HBV DNA and quantitative HBsAg may potentially guide the duration of (peg)interferon treatment as in the case in chronic hepatitis C. One difference between chronic hepatitis B and chronic hepatitis C is the availability of oral nucleos(t)ide analogues, which provide an alternative for poor responders to peginterferon therapy. The HBV DNA roadmap concept can serve as a guide to reduce the emergence of drug resistance, particularly when nucleos(t)ide analogues with a low genetic barrier of resistance are used. The current unsettled question is what to do for suboptimal responders, as the current recommendations are largely based on expert opinion rather than clinical evidence. Nonetheless, predictors of response are useful to provide the most appropriate antiviral therapy to the most suitable patients, in order to achieve the best response and improve the clinical outcome of chronic hepatitis B patients.
No sources of funding were used in the preparation of this article. Henry L.-Y. Chan is a member of the advisory boards of Novartis Pharmaceutics, Bristol-Myers Squibb and Pharmasset. Grace L.-H. Wong has no conflicts of interest that are directly relevant to the contents of this article